Discussion Forum

Can I apply the damage as a point source??

Marina,

In the absence of more detailed information, I'm not sure I'm understanding your problem correctly, but here are a few items that may be relevant:

1) The variable(s) for which you are solving the convection-diffusion problem must be solved for throughout the entire simulation over the entire domain where they may be needed. You can't toggle them on or off during the simulation unless you know beforehand the times at which they must be turned on or off.

2) The convection and diffusion terms can be turned off (or made extremely small) using a smoothed step function, where the step occurs at the damage threshold.

3) Making the convection-diffusion process contingent on the distance to an arbitrarily located damage location is trickier, but it can be done. Either a non-local coupling must be used or a dependent variable must be created to solve for the locations that are affected by nearby damage.

--
Luke Gritter
AltaSim Technologies

Thanks for the reply Luke.

I think I got the problem solved. For now I am defining damage as a constant variable equal to 1 and I only define it in a particular domain of the entire geometry.

Now I have a different problem concerning the diffusion process.
I have my geometry divided in different domains so that I can assign each diffusion process to different parts of the geomtry.
I have an initial concentration of cells in domain A and those cells diffuse along that domain (A) and I also want them to diffuse to the "neighbour" domain B. For that, I create flux between the boundary that separates A and B. I run the study but at the end the concentration has varied along domain A but nothing has appeared on the domain B. I have also tried defining an open boundary condition. I want the initial concentration of these cells to be zero on domain B (or very small).

What am I doing wrong?

Marina,

From your description, it sounds like you have a diffusion interface (and hence a dependent variable) that is only active on domain A, and another (with a different dependent variable) that is only active on domain B. If this is correct, and if you want the concentration to be continuous across the boundaries between domains A and B, you need to apply a Dirichlet boundary condition (i.e. a "Concentration" boundary condition in the diffusion interfaces) at the boundaries between domains A and B. For the variable in domain A, the concentration at these boundaries should be set to be equal to the variable in domain B, and vice versa.

If I'm not correctly understanding your problem, you could upload a model file or provide a more detailed description of the physics setup for the diffusion variables.

--
Luke Gritter
AltaSim Technologies

Hi again, thanks a lot for your fast reply.
I am going to be a little more specific on my problem cause I don't think I am being clear.

I am studying the difussion process along a tissue wall (after damage by a medical device is done) where different cells migrate, proliferate and die and substances are degraded or produced. The fluid part is not implemented yet so there is no convection, just diffusion.

There are two types of cells that diffuse and some substances that degrade or are produced and regulate the proliferation or apoptosis of these two cells species.

The geometry is composed by two diferent layers that are subdivided in other domains.

The endothelium layer is divided in two halfs. One half has no cells and the other half has cells that diffuse and also move to the second "emplty" half (which at t=0 has no cells).
The media layer is also divided in two parts. One part has an initial concentration of other cell species that diffuse around, and also to the emplty part of the media layer and to the endothelium.

I define the difussion of each cell species with two different "transport of diluted species modules" where I define the initial concentration, boundary conditions, diffusion coefficient and reactions.

I also have a differential equations module where I define the substances behaviors and the initial concentrations.

When I run the model the two types of cells that have diffusion only diffuse inside their own domain and the concentrations of the cells and substances varies but the cells do not appear on the neighbour domains where I want them so maybe I am not setting right the boundary conditions for the diffusive species??

I hope I am a bit clearer now.

Thank you very much for your time

Marina,

Each "Transport of Dilute Species" must be active on all domains in which its species may diffuse. To set the different initial concentrations in different domains, use multiple "Initial Values" nodes in each interface. For example, activate a "Transport of Dilute Species" in all four domains (two endothelium domains and two media domains). On the default "Initial Values" node, set the initial concentration for the domain in which the cells begin (by default this "Initial Values" node will apply to all domains until it is overridden). Then add another "Initial Values" node and apply it to the three domains where there is zero initial concentration. You do not need to apply any boundary conditions between the domains. If you want to use different diffusivities for different domains, you can either add additional "Diffusion" nodes or use "Variables" nodes under Definitions to set this up.

--
Luke Gritter
AltaSim Technologies

Thank you very much Luke. That makes a lot of sense. I dind't know it was possible to set two different "initial value" nodes to different domains since I saw that by default you have to apply the initial values to all the domains. I am new in COMSOL and I am learning every day new things that can be done.

I will try that out and hopefully it will work.

Thanks again for your fast and very helpful response.

Marina

Hi Luke,

I just wanted to let you know that setting a different initial concentration for each domain worked. At the beginning it returned a singularity error because the difference between concentrations at each domain was too big, resulting in a too large initial step size for the solver. I used a smoothed step function and it worked fine.

Thank you again.

Marina

Hi, I have a new question.

I want to set a varible (p) equal to 1 at point "A" and I want it to decrease within a domain until it reaches 0. So I want to set pressure dependent on r and z since my model is 2d axisymmetric, and I also want it to decrease with time.

Any thoughts?

Thank you